SCS-CN-INSPIRED RAINFALL-RUNOFF, SEDIMENT YIELD, AND ENVIRONMENTAL FLOW MODELLING

Abstract

Water is not only vital for life but also a natural resource of paramount importance for well-being of the society on the earth, essentially required for growth, sustainability of ecosystem, and prosperity of nation. For sustainability of the ecosystem, water management is required and it can be used for improved watershed models to address the management issues more effectively. It is well known that the watershed models are used to analyze the quantity and quality of stream flow, erosion and sediment yield, reservoir system operations, groundwater development and protection, surface water and groundwater conjunctive use management, water distribution systems, irrigation water use, and a range of such water resources management activities. The component processes of hydrologic cycle over a watershed, viz., precipitation, evapotranspiration, detention, interception, infiltration, percolation, interflow, base flow, overland flow, and runoff etc., depend on climatic and catchment characteristics which vary both spatially and temporally. Thus, for reliable predictions of the runoff and sediment yield from land surface into streams and rivers, it is very important to understand rainfall-runoff-sediment relationship for making reliable estimates of runoff and sediment yield from a watershed. Simultaneously, the hydrological models used should also be parametrically efficient for the available data of watershed. The SCS-CN method has been used by a number of researchers for runoff estimation worldwide since its inception in 1956. As a result, it has been a subject of intense and extensive exploration for its formation, rationality, applicability and extendibility, physical significance, and so on. Beside others, the method still inherits a major structural inconsistency associated with S-CN mapping, where S = potential maximum retention and CN = curve number, and results into abnormality in description of watershed behavior, i.e., complacent, standard, and violent, and runoff estimation based on the existing SCS-CN method. Thus, there is a need to revisit the S-CN mapping relationship for improvement and, in turn, improved SCS-CN methodology. Suspended sediment is another hydrologic variable of severe environmental concern. Its transport in river systems is of paramount importance for ecologists and water and land- managers. The presence of sediment in water greatly influences the design and operation of hydraulic structures, like canals, diversions and dams. Therefore, its accurate assessment can be of potential use in future water resources management policies. This assessment is nonlinear and quite involved in ii nature and depends on a number of factors such as flow, precipitation, topography and soil and land use characteristics of the watershed. A number of models of varied complexity are available to model soil erosion and sediment yield. The Universal Soil Loss Equation (USLE) is the most popular model for soil erosion. Since sediment yield depends on surface runoff, the erosion models are often coupled with rainfall-runoff models, such as the existing SCS-CN methodology, and thus, there exists a scope for further improvements. The present study was thus taken up with the following specific objectives:  To develop an improved SCS-CN model based on the proportional equality revised for the first order linear hypothesis and Horton infiltration concepts for runoff estimation.  To propose CN-P mapping relation based on the physical description of CN for describing the watersheds behavior realistically.  To propose improved conceptual sediment yield models based on coupling of Universal Soil Loss equation (USLE) and modified SCS-CN method.  Prediction of environmental flow condition using runoff curve number. To accomplish the above tasks of more accurate runoff and sediment yield estimation, SCS-CN methodology was revisited in perspective of its various component processes. The proportional equality of this methodology, which is in core of the CN-concept was modified by replacement of the parameter potential maximum retention (S) by Soe-P, where So is the initial (or absolute maximum potential retention) and is the decay parameter. It was founded on the concept that S is in correspondence with P (or Pe = P –Ia), which is variable whereas So being absolute is not variable, and therefore, a better substitute for being a parameter. This modified version is validated using the rainfall and runoff data from Hawkins (1993), Strange (1882), and Tehri catchment (Uttarakhand, India); the last two belong to southern and northern parts of India, respectively. The proposed modification to the application approach of the SCS-CN methodology (i.e. Model 3) is more rational and has the efficacy to describe the watershed behavior more rationally/scientifically and resolve the issue of CN decaying with increasing rainfall (P). Besides, the methodology is found to be consistent when applied to all three datasets. The performance was evaluated using NSE, RMSE and Bias error criteria. Though the proposed methodology generally performed as well as the existing one in runoff estimation, it described realistically the physical behavior of the three types of watersheds, viz. complacent, standard, iii and violent, and the increasing (in contrast to the established decreasing) trend of runoff coefficient with rainfall.. While revisiting the above SCS-CN methodology, it was also investigated for the S-CN mapping. It was revealed that the parameter Curve Number (CN) actually represents the runoff potential for a given amount of rainfall, which at present is considered as 1 inch (= 254 mm). Therefore, in real-world applications, when the rainfall varies or is different from given rainfall amount, is necessary to revise the Curve Number (CN) as CNP, which corresponds to the actual amount of rainfall P. This major structural modification in the existing S-CN mapping relationship was proposed to solve abnormality in watersheds behavior, i.e. complacent, standard, and violent, as described by Hawkins (1993). The proposed modification effectively resolves the abnormalities and supports the general notion that the runoff coefficient (C) (or CN, another form of C) increases with increasing P. This modification also describes the CNP-P relationship and, in turn, the behavior of three watersheds for growing rainfall. The results show that the pre-derived CNP-P relationship for a watershed can be an improved alternative for runoff prediction using SCS-CN methodology. These relationships were also derived for Strange and Tehri catchment data and results showed that both the models performed well on these data sets. The study revealed that proposed modification also showed an enhanced model performance based on NSE, RMSE and Bias error criteria. To meet the third objective, the applicability of USLE-coupled SCS-CN models was examined for computing total sediment yield from a storm event using the rainfall-runoff-sediment yield data observed from 09 experimental watersheds (plot size 12×3 m2) of different land uses, soils, and slopes. The sediment rating curves drawn between the observed sediment and observed runoff for the plots of different slopes showed increasing trend of sediment yield (mass) with runoff and slope for all land uses, viz., maize, finger millet, and fallow land. The rate of increase in sediment yield with runoff was sharper for greater slopes, and vice versa. Both existing and proposed SCS-CN models were tested using observed sediment and runoff data (Chapter 4). The proposed sediment yield models (PS1 with λ = 0.0 and PS3 with varying λ) performed much better than the existing models (S1 with λ = 0.0 and S3 with varying λ), and significantly better than S2 with λ = 0.2 when applied to both years 2016 and 2017 data. The higher slope plots generated higher sediment yield and runoff, and vice versa. The sediment yield models were ranked for performance as: PS3>PS1>PS2. The runoff models PR3 with varying λ and PR1 with λ = 0 performed approximately similar and model PR2 with λ = 0.2 performed poorer than any other proposed models. iv For the accomplishment of the last objective, a relation of percentage of average annual flow (%AAF) with CN has been explored and its application has been demonstrated on seventeen catchments located in different river basins of India. These catchments include nine catchments from Godavari basin (viz. Ashti, Bimini, Bhatpalli, Satrapur, Jagadalpur, Nandgaon, Ramakona Hivra, P.G. Penganga), two catchments of Mahi basin (Chakaliya and Dhariwad), four catchments of Mahanadi basin (viz. Baronda, Basantpur, Ghatora, Rampur), one catchment of Brahmani-Baitarini basin (i.e. Jenepur) and one catchment of narmada basin (i.e. Kogaon) falling in sub-tropical, and sub-humid climatic regions of India have been used. The coefficient of determination of more than 0.6 for most catchments shows the existence of an excellent relationship between CN and %AAF (used to describe EF condition). Hence, the environmental flow condition of these catchments may be determined using CN for known catchment characteristic.